Pulse cyclophosphamide is more effective than prednisone alone in preventing renal failure in lupus nephritis. This study sought to define whether pulse methylprednisolone could equal pulse cyclophosphamide in preserving renal function, and whether there was a difference between long and short courses of pulse cyclophosphamide in preventing exacerbations of lupus. Patients were treated with prednisone and randomized to receive concomitantly (a) pulse methylprednisolone monthly for 6 months, or (b) pulse cyclophosphamide monthly for 6 months, or (c) pulse cyclophosphamide monthly for 6 months followed by a maintenance regimen every 3 months for an additional two years. Patients treated with pulse methylprednisolone had a higher probability of developing renal insufficiency than patients treated with the long course of pulse cyclophosphamide. In addition, patients treated with only a short course of pulse cyclophosphamide had a higher probability of major exacerbations of lupus than those treated with the extended course of pulse cyclophosphamide. Studies of ovarian toxicity in patients treated with pulse cyclophosphamide showed that risk is affected by both age at treatment and total doses of cyclophosphamide. A series of clinical and pathologic factors were analyzed for their ability to predict renal failure subsequent to entry into the therapeutic trials. Black race, age >30 years, anemia, elevated serum creatinine and hypocomplementemia significantly predicated adverse renal outcomes. However, renal biopsy features contributed additional prognostic information and significantly enhanced the strongest clinical model.